MicroPET Imaging of Bacterial Infection with Specically Nitroreductase Responsive 18F-labeled Nitrogen Mustard Analogues

Purpose Bacterial infection and antibiotic-resistant are still serious threats to human health. Here, we aim to develop two novel radiotracers 18 F-NTRP and 18 F-NCRP with specic nitroreductase (NTR) responsive to image the deep-seated bacterial infection with PET for differentiating infection from sterile inammation. 18 F-NTRP and 18 F-NCRP were synthesized via a one-step method. All the steps of tracers radiosynthesis were successfully adapted in the All-In-One automated module. After the physiochemical properties of 18 F-NTRP and 18 F-NCRP were characterized, their specicity and selectivity to NTR were veried in E. coli and S. aureus as well. The ex vivo biodistribution of tracers was evaluated in normal mice. MicroPET-CT imaging was acquired in mouse models of bacterial infection and inammation after the administration of 18 F-NTRP and 18 F-NCRP.


Introduction
Although numerous antibiotics have been developed and used in the treatment of bacterial infections for near a century, bacterial infections and multidrug-resistant bacteria are still pressing public health concerns [1,2]. Increasing numbers of multidrug-resistant bacteria may be due to the overuse of antibiotics and increased potential risks from invasive surgery or therapy processes [3]. For decreasing the unnecessary usage of antibiotic therapy, a critical issue is differentiating bacterial infection from sterile in ammation accurately, because the treatments for infection and sterile in ammation are diametrically different in the clinic. However, rapid and accurate technologies for early detection and localization of infections in the patient remain a challenge. Traditional tools such as microscopy based on culturing of tissue biopsies or blood samples are not accurate enough, because of the sampling error/bias, furthermore, such methods are most suitable for the later stage of infection [4]. Therefore, noninvasive imaging methods are necessary, and structural imaging techniques such as X-ray, computed tomography (CT), magnetic resonance imaging (MRI) have been applied in infection diagnosis. These techniques are based on the presence of abnormal uid that occurred late within tissues. However, the abnormal uid may also appear under but sterile in ammatory conditions [5]. Therefore, these techniques are nonspeci c, cannot differentiate infections from in ammation [6].
However, there are also limitations of these tracers, such as low speci city in patients, limited differentiating ability of particular bacterial strains and dependent on bacterial metabolic state for some tracers. Thus, it is important to exploit potential targets of bacteria for infection imaging.
In this study, bacterial nitroreductase (NTR) was selected as a new target for infection imaging. NTR is a family of avin-containing enzymes, widely exist in most Gram-positive, Gram-negative bacteria and hypoxia tumors [26,27]. Aromatic nitro groups can be reduced to nitroso group, hydroxylamine group, and the amino group ultimately under the catalysis of NTR [28]. In this reductive reaction, reduced nicotinamide adenine dinucleotide (NADH) or nicotinamide adenine dinucleotide phosphate (NADPH) is a vital cofactor, provide electrons in the reduction process [29]. Notably, NTR is also recognized as a biomarker of bacterial infection in clinical, food, and environmental samples. Up to now, NTR-activated imaging tools have attracted great attention for infection imaging, include electrochemical and uorescent imaging [30][31][32]. Although NTR activated optical imaging probes used in infection imaging have already been well described and studied, radiotracers responsive to bacterial NTR for detecting and localizing deep-seated infections in vivo are still scarce.
Herein, we designed and radiolabeled two NTR activated tracers, nitrogen mustard analogues decorated with aromatic nitro groups ( 18 F-NTRP and 18 F-NCRP), as new PET tracers applied for infection imaging. 18 F-NTRP and 18 F-NCRP are consists of two separate functional domains: an e cient NTR responsive moiety "aromatic nitro group" and a potential trapped effector "nitrogen mustard". Nitro is a kind of strong electro-withdrawing group, decreases the electron density of nitrogen mustard required for alkylation. In bacterial infection sites, the nitro group of 18 F-NTRP or 18 F-NCRP is likely to be reduced to electron-donating group amino with NTR speci cally. After that, the lone pair electrons developed on nitrogen mustard are conducive to the formation of a highly electrophilic aziridinium ring, which enables cross-link to bacterial DNA through covalent bond.
In this study, we developed fully automated radiosynthesis methods of 18 F-NTRP and 18 F-NCRP, which consistently yielded the nal products in large quantity with high radiochemical purity (RCP) and molar activity. Both 18 F-NTRP and 18 F-NCRP were speci cally reduced by NTR and readily taken up and retained by bacteria. Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus) were chosen as representative Gram-negative and Gram-positive bacteria and typically overexpress NTR [33]. PET imaging in E. coli and S. aureus infected mice models were also performed with 18 F-NTRP and 18 F-NCRP. In addition, many frequently encountered infections are caused by E. coli and S. aureus, such as pneumonia, diarrhea, superinfection, and osteomyelitis, etc. [34,35]. We also performed PET imaging noninvasively differentiate infection in vivo from sterile in ammation with 18 F-NTRP and 18 F-NCRP. Therefore, we think 18 F-NTRP and 18 F-NCRP having outstanding performance in early detecting, and speci cally locating infections in vivo.

Materials And Methods
Detailed materials and methods for synthesis, characterization, in vitro and in vivo evaluations of tracers were described in the ESI section.

Results And Discussion
Radiochemical Synthesis of 18 F-NTRP and 18 F-NCRP We designed and synthesized two PET tracers, 18 F-NTRP and 18 F-NCRP, for infection imaging in vivo. The overall rationale of 18 F-NTRP and 18 F-NCRP for infection imaging is that radiotracers enable the response to bacterial NTR speci cally and then be trapped in living bacteria. The structure of tracers is aryl nitrogen mustard decorated with a para-NTR-activated nitro aromatic group (Scheme 1). Electron-withdrawing nitro group decreases the electron density of the benzene ring e ciently and makes the lone pair electrons of the nitrogen mustard delocalized to nitro. Thereby, impeding the formation of an electrophilic aziridinium ring and resulting in the nitrogen mustard group being unable to connect to DNA. However, in the infection site, the radiotracer is speci cally reduced by bacterial NTR, the nitro transforms into electron-donating amino, which triggers an increase of electron density to a mustard moiety, and promotes the formation of aziridinium ring. The electrophilic aziridinium ring is prone to cross-link to bacterial DNA through a covalent bond, resulting in radioactivity being trapped in bacteria. Herein, we synthesized two PET tracers of 18 F-NTRP and 18 F-NCRP with different labeling positions and nitrogen mustard groups to investigate the structure-function relationship (Scheme 1).
The synthetic strategy was devised to synthesize the precursors and reference compounds of 18 F-NTRP and 18 F-NCRP according to the routes depicted in Figures S1 and S2 with high yields. Detailed synthetic procedures and full characterization data of intermediates, precursors, and reference standards can be found in the ESI section. 18 F-NTRP was synthesized with a tosylate precursor via a one-step method and 18 F-NCRP was radiolabeled with a nitro group precursor at a mild situation following the literature [36]. All the steps of tracer radiosynthesis were achieved in 90 -110 min, including the initial receiving, eluting, and drying of [ 18 F] uoride that from cyclotron target water, probes radiolabeling via nucleophilic substitution reaction, product puri cation, and injections formation were successfully adapted in the All-In-One automated module ( Figures S3 and S4). Final clear injections were obtained with a pH of 6 -7.
Starting from 14.8 -25.9 GBq of [ 18 F] uoride, the overall decay-uncorrected radiochemical yields (RCY) were 20.24 ± 6.25 % (n = 13) for 18 F-NTRP and 11.3 ± 3.78 % (n = 12) for 18 F-NCRP respectively. Therefore, robust and fully automated radiosynthesis methods have been developed, which can be readily scaled up for the production of multiple clinical doses. The molar activities of 18 F-NTRP and 18 F-NCRP were 320 ± 40 GBq/µmol (n = 13) and 275 ± 33 GBq/µmol (n = 12) at the end of synthesis (EOS), with high RCP (over 95%). The chemical identities of 18 F-NTRP and 18 F-NCRP were con rmed by coinjection of the nal products with non-radioactive reference standard 19 F-NTRP and 19 F-NCRP for radio-HPLC analysis. The consistent HPLC retention time of the radioactive tracers and their corresponding non-radioactive reference standards ( Figure 1A) suggested that the chemical structures of 18 F-NTRP and 18 F-NCRP were correct. In addition, both 18 F-NTRP and 18 F-NCRP showed excellent stability in vitro after 2 h incubation in saline or serum ( Figures 1B, 1C). Taken together, these newly developed 18 F-NTRP and 18 F-NCRP tracers were obtained with high RCP, molar activity and excellent stability consistently, which satis ed the requirement of potential clinical applications.
Speci city and Selectivity to NTR Since 18 F-NTRP and 18 F-NCRP were designed to a speci c response to bacterial NTR, we next investigated the speci city and selectivity of the tracers to NTR. We rst studied the speci city of 18 F-NTRP and 18 F-NCRP to NTR in a Tris-buffered solution. The amino-substituted species 19 F-NTRPH and 19 F-NCRPH were synthesized as the reference standard to analyze the reduced productions (Figures S1, S2). After reduction, we found that in the presence of NTR and NADPH, 88 % of 18 F-NTRP have been reduced, and 37 % of reduced products were 18 F-NTRPH (Figure 2A). Similar results were observed in 18 F-NCRP and NTR solution, in which 89 % of 18 F-NCRP have been reduced, and 17 % was translated to 18 F-NCRPH ( Figure 2B). These results demonstrated that 18 F-NTRP and 18 F-NCRP were speci cally responsive to NTR in vitro. Even though most of the parent tracers had been reduced, the transformation rate of 18 F-NTRP to 18 F-NTRPH and 18 F-NCRP to 18 F-NCRPH were relatively low. We speculate that the low transformation rate may be due to a portion of the reduced products in the solution were oxidized to the corresponding parent compound or intermediates under the atmosphere. Additionally, we also carried out the selective study of 18 F-NTRP and 18 F-NCRP to various species. We compared the transformation rate of 18 F-NTRP to 18 F-NTRPH and 18 F-NCRP to 18 F-NCRPH in different kinds of small molecule solutions. As shown in Figures 2C and 2D, 18 F-NTRP and 18 F-NCRP displayed an excellent selectivity towards NTR. The transformation rate of 18 F-NTRP to 18 F-NTRPH and 18 F-NCRP to 18 F-NCRPH incubated in numerous species was about 0.7 -1.3 %, similar to that observed in saline. However, in the presence of NTR and NADPH, the transformation rate was 13 -28-fold higher than achieved with other species, even though the concentration of NTR was much lower than with other species (0.2 mM vs 50 mM). These results indicated that 18 F-NTRP and 18 F-NCRP have good speci city to NTR and exhibited excellent antiinterference ability.

Bacterial Uptake Studies
We subsequently evaluated the bacterial uptake and retention of 18 F-NTRP and 18 F-NCRP in representative human pathogens E. coli and S. aureus in vitro to determine whether tracers could be internalized and trapped in bacteria. As shown in Figures 3A and 3B, obviously much higher uptakes of 18 F-NTRP and 18 F-NCRP with both living Gram-negative (E. coli) and Gram-positive (S. aureus) bacteria were observed than that with heat-killed bacteria. Moreover, 18 F-NTRP showed increased accumulation in living bacteria with time, the uptakes reached 3.64 ± 0.46 % in E. coli and 2.76 ± 0.23 % in S. aureus at 90 min, which were about 5 -6-fold higher than the controls (0.62 ± 0.11 % and 0.55 ± 0.18 % in heat-killed E. coli and S. aureus, respectively). In addition, E. coli exhibited a 1.3-fold uptake of that achieved by S. aureus. Possibly due to the higher levels of NTR expressed in E. coli than in S. aureus. Compared to 18 F-NTRP, 18 F-NCRP demonstrated higher incorporation in both living E. coli and S. aureus (7.71 ± 0.73 % and 6.13 ± 0.18 % respectively), which is about a 2-fold higher than seen with 18 F-NTRP. The discrepancy of cellular uptake between 18 F-NTRP and 18 F-NCRP may be due to the 18 F-NCRP being able to be incorporated into DNA more rmly because there are two alkyl chains in 18 F-NCRP that can cross-link with bacterial DNA, but only one alkyl chain in 18 F-NTRP.
Based on the higher uptake of 18 F-NCRP in E. coli, we further investigated the retention capacity of 18 F-NCRP in E. coli. Results in Figure 3C showed that 18 F-NCRP exhibited good retention in bacteria. After incubation for 10, 20, and 30 min, there were 81.43 ± 21.03 %, 70.23 ± 12.63 %, and 66.41 ± 16.96 % radioactivity were retained in E. coli, respectively.  Figure 4A demonstrated that radioactivity derived from 18 F-NTRP was well accumulated in both S. aureus and E. coli infected muscles. Signi cantly, the uptake was 2.4 ± 0.2 %ID/g in E. coli infected muscles, which was 1.47-fold to that of S. aureus infected muscles (1.63 ± 0.15 %ID/g) and 4-fold to the uninfected control muscles (0.65 ± 0.09 %ID/g) ( Figure 4B). The good target-to-control contrast enabled 18 F-NTRP to observe infections in deep sites. In addition, the intravital radioactivity uptake by bacteria in infected mice was well consistent with in vitro uptake by bacterial culture. We further studied the biodistribution of the radioactivity in infected triceps, both bilateral triceps of mice were excised after PET imaging and followed by autoradiography. Figure S5A showed that obviously infection tissues were observed in bilateral triceps. Moreover, the radioactivity in triceps was mostly accumulated in infection areas and radioactivity in E. coli infected tissues was higher than in S. aureus infected samples ( Figure S5B). In the 18 F-NCRP group, PET imaging ( Figure 4A) showed that visible accumulation of radioactivity was observed in E. coli and S. aureus infected muscles. Moreover, low radioactivity uptake was observed in blood, liver, lung, heart, and brain except for the gall bladder, making 18 F-NCRP more favorable for a localized infection diagnosis. PET images of 18 F-NCRP demonstrated that the radioactivity in E. coli infection was 4.05 ± 0.49 %ID/g, 1.48fold to that observed in S. aureus infection (2.7 ± 0.3 %ID/g), and 3.2-fold to uninfected control muscles (1.24 ± 0.26 %ID/g). Compared to 18 F-NTRP, 18 F-NCRP exhibited better properties in bacterial imaging.
The uptakes of 18 F-NCRP in E. coli and S. aureus infected tissues were 1.67 and 1.65 times higher than that of 18 F-NTRP, respectively ( Figure 4B). Signi cantly, the fast clearance of 18 F-NCRP from healthy organs or tissues, which helps generate a high infection-to-background ratio and provides bene ts for clinic infection diagnosis. In addition, we found that E. coli infected triceps was much more severe than S. aureus infected ones, and autoradiography results showed that the radioactivity in E. coli infected muscle was also higher than that in S. aureus infected obviously ( Figure S5C, S5D). Thus, 18 F-NCRP imaging could distinguish the severity of infection in vivo.
The ability to differentiate active infection from sterile in ammation is a critical requirement for bacteriaspeci c radiotracers. Hence, we developed a mixed infection and in ammation mice model for PET imaging with 18 F-NTRP and 18 F-NCRP. We rst evaluated the infection and in ammation model with 2-[ 18 F]-Fluoro-2-deoxy-D -glucose ( 18 F-FDG), which is commonly used in clinical diagnosis. PET imaging of 18 F-FDG showed that the uptakes in in amed and infected muscle were 15.47 ± 2.46 %ID/g and 6.70 ± 0.34 %ID/g, respectively ( Figure S6). Even though the uptakes of infected and in amed areas were much higher than normal muscle (2.78 ± 0.77 %ID/g), there is no su cient discrepancy between infected and in amed muscles, indicating that 18 F-FDG could not differentiate bacterial infection from sterile in ammation effectively in vivo. PET imaging with 18 F-NTRP or 18 F-NCRP was also performed in the same model mice. At 40 min p.i., apparent 18 F-NTRP radioactivity was accumulated in E. coli infected muscle (1.90 ± 0.30 %ID/g), which is 3-fold to that accumulated in in amed muscle and 3.8-fold to that seen in healthy control muscle. However, the uptakes between in amed muscle and healthy control muscle were no signi cant difference (0.67 ± 0.08 %ID/g vs 0.48± 0.07 %ID/g). In the 18 F-NCRP group, the uptake ratio of E. coli infected muscles to in ammation muscles was 2.5 (3.90 ± 0.30 %ID/g vs 1.52 ± 0.17 %ID/g). Moreover, E. coli infection uptake was 3.8-fold to healthy muscle uptake (3.90 ± 0.30 %ID/g vs 1.02 ± 0.22 %ID/g, Figures 4C, 4D). The low uptake in in ammation muscles was in accordance with the cellular uptake results of 18 F-NTRP and 18 F-NCRP in heated-killed bacteria. That is due to the complete Freund's adjuvant is consisted with heat-killed M. tuberculosis to inducing in ammation.

Metabolic Analysis
Next, we analyzed the metabolic stability of 18 F-NTRP and 18 F-NCRP in vivo. Healthy mice were injected with 18 F-NTRP or 18 F-NCRP (37 -38 MBq) via tail vein. Blood and urine samples were collected at 2, 10, 30, or 60 min p.i., respectively. Collected samples were preconditioned to remove the biomacromolecules and then were subjected to radio-HPLC analysis. As shown in Figure 5A, for the blood samples, more than 95 % of intact 18 F-NTRP was observed with a retention time of 8.24 min at 2 min p.i. At 10 min p.i., about 73.3 % intact 18 F-NTRP remained and a small radio-metabolite was detected at 2.43 min on HPLC chromatogram. Moreover, less intact 18 F-NTRP was found at 30 min p.i. (40.8 %) and nally no intact 18 F-NTRP left at 60 min p.i. At the same time, the radio-metabolite peaked at 2.43 min was increased to 97.51 % from 10 min to 60 min p.i. For the urine samples, no radio-metabolite was detected at the rst (2 min p.i.). After that, no intact 18 F-NTRP was observed, but a metabolite was observed at 2.69 min with gradually increased concentration from 10 -60 min p.i. Due to the minor de uorination being observed in the following biodistribution study, we suspect that the radio-metabolite in blood and urine consisted of a little 18 F-uoride and electrophilic aziridinium ring derived from 18 F-NTRP. In addition, NTR existing in normal organs promotes the formation of electrophilic aziridinium rings from 18 F-NTRP. Compared to 18 F-NTRP, 18 F-NCRP was cleared from blood more quickly. The percentage of intact 18 F-NCRP in the blood sample was decreased to 46.46 % at 2 min p.i., and three metabolites were detected at 3.95 min, 6.82 min, and 9.91 min, respectively. We speculate that these metabolites were three different intermediates of the aromatic nitro group reduced to amino by NTR in vivo. From 10 -30 min p.i., only about 5.63 % intact 18 F-NCRP remained in the blood sample and the proportion of radio-metabolite at 3.95 min on HPLC chromatogram was increased to 88.96 %. In the urine sample, an apparent radio-metabolite was detected at 3.67 min on the HPLC chromatogram. In addition, only 0.2 % of parent 18 F-NCRP was detected from 2 -10 min p.i., but it was undetectable after 30 min p.i. (Figure 5B). Finally, we studied the biodistribution of 18 F-NTRP and 18 F-NCRP in normal mice. Healthy mice were injected with 18 F-NTRP or 18 F-NCRP (0.37 MBq/mouse) vial tail vein, followed by sacri ced at desired time points. At 2 min p.i. of 18 F-NTRP, high uptakes were observed in NTR expressed tissues, such as liver (15.21 ± 1.20 %ID/g), kidney (16.56 ± 2.22 %ID/g). Most of the rest radioactivity were concentrated in blood (6.02 ± 0.62 %ID/g), heart (7.41 ± 0.64 %ID/g), lung (12.38 ± 1.63 %ID/g), spleen (6.74 ± 0.99 %ID/g), brain (6.80 ± 0.66 %ID/g), and intestines (7.61 ± 0.10 %ID/g). Less radioactivity was accumulated in the stomach (3.36 ± 0.26 %ID/g), bone (2.79 ± 0.41 %ID/g), and muscle (4.88 ± 0.55 %ID/g). At 10 -60 min p.i., radioactivity was gradually cleared from most organs. Lower muscle uptake is well supporting the muscle infection imaging. However, incorporated radioactivity in the kidney and intestines was increased to 17.33 ± 2.94 %ID/g and 8.61 ± 1.56 %ID/g, which might be due to the main renal and intestines excretion of 18 F-NTRP. Notably, we found signi cant radioactivity in the small and large bowel, especially concentrated in excreta. This background uptake was a result of the rich microorganisms in the enteric canal. Low bone uptakes from 2 min to 60 min p.i. may indicate extremely low de uorination of 18 F-NTRP in vivo ( Figure 6A). For 18 F-NCRP, the radioactivity distribution patterns were similar to those observed with 18 F-NTRP. At 2 min p.i., high uptakes in blood (7.96 ± 0.32 %ID/g), heart (9.24 ± 0.43 %ID/g), lung (14.00 ± 0.39 %ID/g), spleen (12.65 ± 0.63 %ID/g), brain (8.45 ± 0.17 %ID/g), liver (18.80 ± 0.60 %ID/g), and kidneys (12.72 ± 0.26 %ID/g) were found. Radioactivity derived from 18 F-NCRP was cleared quickly from most organs except that kidney uptake was increase to 16.12 ± 0.42 %ID/g at 10 min p.i., followed by decreased to 2.34 ± 0.37 %ID/g at 60 min p.i. Moreover, uptakes in the bone, stomach and intestinal were increased to 11.53 ± 1.22 %ID/g, 9.40 ± 0.23 %ID/g and 5.74 ± 0.49 %ID/g, respectively at 60 min p.i., which is higher than that observed at 2 min p.i. These results indicated that 18 F-NCRP had slight de uorination in vivo and was mainly excreted by the kidneys (Figure 6B).

Conclusions
In this study, we developed NTR-speci c responsive tracers 18 F-NTRP and 18 F-NCRP for PET imaging of deep-seated bacterial infection. Automated radiosynthesis of tracers were consistently obtained with high RCY, RCP and molar activity. Both tracers displayed high sensitivity NTR response and could be selectively accumulated in living bacteria and bacteria-infected tissue but not in heat-killed organisms. In     Uptakes of 18F-NTRP and 18F-NCRP in the infected and in amed triceps were derived from PET-CT images.